The Liberty Daily 
Humans First chair Amy Kremer speaks with a reporter during the Data Center Revolt Tour. At stops across the country, residents and reporters have asked many of the same questions: Who is behind these projects, what is actually known, and what remains a projection?
In the last installment, I introduced you to Janet in Iowa, Carol and Ben in Kansas, Ana in Pennsylvania, families in Virginia, and small-town residents across the country wondering what will become of the places they have spent their lives building and preserving.
Those stories were about the people living beside America’s rapidly expanding data-center buildout.
But as the bus kept moving, another question followed us from town to town:
How does this whole thing actually work?
America’s race to build the infrastructure behind artificial intelligence is reaching far beyond Silicon Valley. Massive data-center campuses are being proposed in rural towns, farming communities, and small counties that never imagined themselves becoming part of the AI economy.
These facilities are the physical backbone of cloud computing and artificial intelligence. They require land, enormous amounts of electricity, transmission infrastructure, cooling capacity, fiber connections, and — depending on the design — substantial water. They also require extraordinary amounts of capital.
And they are being planned at breathtaking speed.
That speed can make the process look almost incomprehensible from the outside. A community may see drawings of multiple buildings, projected tax revenue, power requests, and water estimates — while still being told that the final customer has not been identified.
That is not necessarily a contradiction.
It is a clue to how this industry actually works.
First: What Is a Data-Center Developer Actually Developing?
Most people picture a simple sequence: a technology company decides it needs a data center, hires a developer, chooses a site, builds the facility, and moves in.
Sometimes that happens.
But another model starts with the site itself.
A developer searches for land that can realistically become a data-center campus. Increasingly, the most valuable ingredient is not simply the dirt.
It is deliverable power.
The developer may buy the land or place it under option. Then it begins trying to eliminate the problems that could make the property unattractive to a future customer: zoning uncertainty, insufficient power, lack of fiber, water and cooling questions, road access, environmental permitting, engineering, and construction feasibility.
Every one of those problems that gets solved can make the site more valuable.
In other words, the developer may not yet be building for a particular company.
It may be building a platform that a future company can use.
The developer can know the envelope before it knows the occupant.
That envelope may include the acreage, approximate number and size of buildings, maximum electrical capacity, substation locations, utility corridors, setbacks, roads, buffers, and a general development schedule.
The eventual tenant may determine many of the details inside that envelope: rack density, chip architecture, cooling technology, actual water demand, electrical configuration, interior fit-out, and how quickly each phase is occupied.
That is why a community can be staring at a detailed site plan while the developer says the final end user is still unknown.
It does not answer every concern.
But it does explain the business model.
How a Farm Becomes a Data-Center Campus
The process gets easier to understand when we break it into steps.
First comes site control: a purchase, an option, or another agreement giving the developer the right to pursue the property.
Then comes feasibility: power, transmission, fiber, cooling, water availability where needed, environmental conditions, road access, and engineering.
Then come the public decisions: rezoning, special-use permits, annexation in some communities, utility agreements, development agreements, tax incentives, and permits.
As those risks are reduced, financing becomes easier to obtain.
Only after enough of those pieces are in place may the site be marketed to, leased by, or custom-built for a hyperscaler, cloud provider, AI company, or another large computing customer.
That explains why local land-use approval can be so valuable.
A parcel of rural land is one thing.
A parcel with zoning, a credible power path, fiber access, and a buildable campus plan is something very different.
That is the canvas developers are trying to create.
Residents, reporters, local officials, and the Humans First team gather at a Data Center Revolt Tour stop. The public often enters the process after site control, power studies, and preliminary planning are already underway.
Why the Frenzy Feels Like the Wild West
The more I learned, the more the energy around this boom reminded me of something I had seen before.
I lived in Oklahoma when the medical-marijuana market opened.
It felt like the Wild West.
A new market. Huge opportunity. Young and hungry entrepreneurs. Changing rules. And a powerful incentive to get there before everyone else.
I see some of that same energy here.
Power is scarce.
Large tracts of land with realistic access to that power are scarce.
Fiber matters.
Permitted sites matter.
And a site that has solved those problems can become far more valuable.
So developers race to secure the ingredients before somebody else does.
That does not make a developer dishonest.
It does mean the public should understand the environment in which the deal is being made.
Some sites will become enormous operating campuses. Some will change hands. Some will be built in phases. Some may be downsized. Some may never reach full buildout. Some may never be built at all.
That is where speculation enters: capital and land are being committed today based on expectations about what computing demand will become tomorrow.
What the Projections Actually Mean
This is where communities can get lost in the numbers.
Projected jobs are not the same as people already employed.
Projected tax revenue is not tax revenue already collected.
Projected water demand is not measured operating consumption.
Projected electrical load is not a customer already purchasing that power.
Projected investment is not the same thing as public revenue.
Those numbers may be perfectly reasonable estimates. They may ultimately prove conservative.
But they are still estimates until the project is built, equipped, occupied, and operating.
That distinction matters because communities are often being asked to make permanent land-use decisions before those outcomes can possibly be known.
A projection is useful information. It becomes dangerous only when people forget that it is a projection.
Why a Billion-Dollar Project Does Not Mean a Billion-Dollar Payday
Take Montgomery County, Missouri.
Amazon announced a $10 billion AWS data-center investment there, with state officials saying the project is expected to create about 400 direct jobs and substantial new property-tax revenue over time.
That kind of number can sound transformative to a small community.
And it can be.
But a $10 billion investment does not mean $10 billion flows into city hall.
Public benefit depends on what becomes taxable, what receives exemptions, how assets are valued and depreciated, which incentives are negotiated, what infrastructure the company pays for, which taxing districts receive the revenue, and whether the entire project is ultimately built.
Data centers can create very real local tax revenue.
But the headline investment number and the community’s recurring benefit are not the same thing.
Hope is not a revenue stream, and an announcement is not a tax payment.
For small towns and rural counties, there can also be a serious imbalance in negotiating resources.
A developer may arrive with attorneys, engineers, utility specialists, financial advisers, site-selection experts, and economic-development teams.
The community may have a small staff, a handful of elected officials, and residents trying to learn an unfamiliar industry in real time.
The negotiating table is not automatically level.
What Happens When Projections Meet Reality?
None of this means every projection is inflated or that every promise will fail.
It means we already have enough real-world experience to know why communities should verify the numbers before making permanent decisions.
One of the clearest examples is unfolding in Spartanburg County, South Carolina.
The Valara data-center project is being built on South Pine Street. It is a massive project — nearly $3 billion when combined with its planned power infrastructure — and it has become the subject of intense public debate.
But the lesson I want people to understand is not simply that residents are upset.
It is what happened to the water projection.
Earlier public information referenced approximately 459,000 gallons of water per day.
By September 2026, NorthMark Compute and Cloud said projected water demand at full buildout would average about 1.2 million gallons per day, with a potential maximum of 1.6 million gallons per day.
The company said the earlier 459,000-gallon figure came from a Phase I wastewater survey and had not been intended to represent the water demand of the completed project.
That is an important distinction.
It may be technically explainable.
But it is also exactly why communities cannot treat early numbers as permanent facts.
A public-works commissioner responded that she was concerned by the magnitude of the change and wanted greater clarity about how the projection evolved and what long-term demand could mean.
If the engineering changes, the projections can change too.
And when they do, the public should know immediately.
| The Valara data center under construction in Spartanburg County, South Carolina. Photo: Ben Gray / Southern Environmental Law Center. |
Spartanburg water figures: WYFF 4, Sept. 21, 2026. Valara construction context: Southern Environmental Law Center, Aug. 26, 2026.
That is the kind of cautionary tale I mean.
Not: “See? Data centers are bad.”
But: “See? Numbers change. Verify them.”
If a project promises jobs, count the jobs.
If it projects tax revenue, publish the actual revenue.
If it estimates water use, meter the water and publish the result.
If it forecasts electrical demand, compare the forecast with the load that actually materializes.
If the engineering changes, update the public.
And if a company or government makes a binding commitment, put it in writing and make sure the people can see whether it was kept.
That is how trust is earned.
Why Secrecy Happens — and Why It Matters
This is one of the hardest parts of the story.
Companies and economic-development organizations often argue that early-stage confidentiality protects site-selection strategy, customer information, financing discussions, trade secrets, and proprietary information while companies compare locations.
That explains why confidentiality exists.
It does not resolve the public-interest question.
When decisions involve public incentives, zoning authority, water, power, infrastructure, or long-term land-use changes, public officials are no longer dealing only with a private business transaction.
They are exercising public authority.
My view is simple:
A public official represents the people — not the developer.
Legitimate trade secrets can be protected without leaving the public blind about material decisions being made in its name.
And timing matters.
Transparency that arrives after a rezoning, incentive package, or infrastructure commitment has already been approved is not the same thing as transparency that allows citizens to participate before the decision is made.
At a Data Center Revolt Tour stop, a state legislator answers community questions and discusses possible legislative responses. For local residents, transparency matters most when it arrives early enough to shape the decision.
We Just Want the Truth
After everything I have seen, I keep coming back to something remarkably simple.
Tell people the truth.
If the jobs are projections, call them projections.
If tax revenue depends on full buildout, say so.
If the customer is not signed, say so.
If water usage depends on a cooling system that has not been selected yet, explain that.
If a power request represents maximum potential demand rather than a committed load, make that clear.
If a company says it will pay for dedicated infrastructure, put that commitment in writing.
If the public could eventually carry other costs, explain how.
And if nobody knows the answer yet, say:
“We don’t know yet.”
The people do not need certainty where certainty does not exist.
They need honesty about the uncertainty.
What destroys trust is not an honest unknown.
It is presenting an assumption as fact, a forecast as a promise, or a private negotiation as though the public has no stake in the outcome.
We are not asking anyone to manufacture bad news.
We are not asking anyone to manufacture good news.
We just want the truth — early enough for it to matter.
What a Community Should Know Before It Votes
By the time a local board reaches a final land-use decision, residents should understand the basic structure of the deal.
Who controls the land?
Who is the developer?
Is there a signed customer?
What is the maximum planned electrical load, and what is actually committed?
What cooling system is assumed in the water estimate?
Which infrastructure costs are contractually assigned to the developer or customer?
What public incentives are available?
Which projected benefits depend on full buildout?
Which commitments survive a sale or change of tenant?
Those are not trick questions.
They are ordinary due-diligence questions.
They are the same kinds of questions an investor would ask before putting money into a project.
A community is putting something valuable into the deal too:
its land-use authority, its infrastructure, its tax structure, and its future.
What Comes Next
Once I understood how the machine worked, another question became unavoidable:
Who is actually financing it?
Because behind the land deals and zoning applications is another world entirely:
Infrastructure funds.
Pension money.
Private credit.
Bonds.
Global asset managers.
And enormous pools of capital betting on what AI demand will become.
That is where the next installment goes.
From the county courthouse to Wall Street.
We just want the truth.
And sometimes the clearest path to the truth is still the oldest one:
Follow the money.
The Data Center Revolt Tour continues across America. The next stop in the investigation is the money behind the machine.
Brooke McGowan is a longtime grassroots activist, former congressional candidate, and former Tea Party News Network staff contributor. She serves as North Carolina State Director for Humans First and travels nationwide with the Data Center Revolt Bus Tour, documenting the human and economic questions surrounding America’s rapidly expanding AI infrastructure.
